The Study Of Audio Classification Based On Wavelet And Support Vector Machine

Audio information processing plays an important role in multimedia applications. Raw audio data is non-semantic and non-structured binary stream, how to extract the semantic content and structure information from raw audio is crucial to deeper processing of audio information, content-based audio retrieval and video parsing with audio assistance. As the core technology of audio structuring, content-based audio classification is a current studied hotspot of audio content automatic analysis.Support Vector Machines (SVM) is proposed in the recent decade as a machine learning theory. As a new machine learning method, SVM shows many good properties over other methods in solving limited samples problems, non-linear and high dimension pattern recognition problems, so it is becoming a new study hot area after neural network. This paper focuses on the two key points of audio classification: feature analysis and extraction, classification algorithm, Main aspects of the paper as follows:(1)We introduce the basic theory of SVM, and the SVM is used to deal with the multi-class audio classification.(2)In this paper, audio features are analyzed in frame level, six environmental sound classes are considered in this paper: the sound of vehicle, bell, wind, ice, machine tool and rain. The experimental results show that the SVM (One Against All) is excellent for environmental audio classification, and the optimal classification accuracy is up to 97.73%.(3)Then we introduce the basic theory of wavelet, and a method that collaborates wavelet analysis and Fourier analysis is used to extract audio features. Sub-band energy and Sub-band ZCR are extracted by wavelet, MFCC(Mel-Frequency Cepstral Coefficients) by Fourier. SVM is used to classified six audio, experimental results show that the method is efficient for audio feature extraction, classification accuracy better than frame level.